Clinical Drug Investigation

, Volume 23, Issue 2, pp 119–128

Influence of Drugs, Demographics and Medical History on Hospital Readmission of Elderly Patients

A Predictive Model

Authors

  • E. F. Ruth Morrissey
    • School of Pharmacy, The Queen’s University of Belfast
    • School of Pharmacy, The Queen’s University of Belfast
  • Michael Scott
    • Academic Practice UnitAntrim Area Hospital
  • Brian J. McConnell
    • Academic Practice UnitAntrim Area Hospital
Original Research Article

DOI: 10.2165/00044011-200323020-00005

Cite this article as:
Morrissey, E.F.R., McElnay, J.C., Scott, M. et al. Clin. Drug Invest. (2003) 23: 119. doi:10.2165/00044011-200323020-00005

Abstract

Objective

To investigate factors that influence hospital readmissions of elderly patients and to construct a robust hospital readmissions predictive model.

Design

Each hospitalised patient was interviewed and medical, demographic and socioeconomic data were obtained from their medical charts. These patients were followed up prospectively for 1 year post discharge with all unplanned readmissions to general medical (including cardiology) wards during this time period recorded. Univariate analysis (chi-squared) was used to identify variables that had at least a minimal association with one or more unplanned readmissions 12 months post discharge (p < 0.25). These were entered into backward stepwise elimination logistic regression analysis (model entry set at p = 0.05). Examination of the significance of the log-likelihood ratio test for each variable determined its contribution to the model.

Patients

Data from a total of 487 elderly patients (≥65 years of age) with non-elective admissions to general medicine wards were used to refine the readmissions model. The refined model was then validated using similar data retrospectively collected from medical charts regarding 732 elderly patients (≥65 years of age).

Results

Multivariate logistic regression analysis yielded a nine-variable model, which contained both predictive and protective variables for one or more readmissions 12 months post discharge from hospital. This model had a specificity of 79.3%, sensitivity of 60.0% and overall accuracy of 71.2% (cut-off point of p = 0.5). When the model was applied to the validation population, an overall percentage accuracy in classification of 65.2% was obtained.

Conclusions

This refined and validated hospital readmissions predictive model could be used by healthcare professionals to help identify vulnerable patients upon admission to hospital, and to put in place a comprehensive discharge planning process.

Copyright information

© Adis International Limited 2003